Wage inequality poised to grow in 2002

2001:3 | QWES Wage Supplement

As the U.S. economy moved from boom to recession, unemployment rose sharply, payrolls fell, and the overall economy began to contract. Little attention, however, has been given to how wages have been affected by the slowdown. An examination of key wage trends through the third quarter of 2001 show the recession’s effect on low-, middle-, and high-wage earners.

To examine recent trends, we look at annualized growth rate of hourly wages each quarter.1 These data represent how quickly wages grew over the latter part of the recovery, and how their growth has responded to rising unemployment and slower growth in recent quarters.

These data show that inflation-adjusted wages continued to grow for workers throughout the wage scale in the third quarter of 2001. This is not unexpected, as wages are typically less responsive to slow growth indicators such as unemployment and job growth. However, wage growth has clearly flattened in the last few quarters. In fact, the wage growth of low-wage workers has slowed considerably, and the growth of wage inequality, after pausing for a few years, is poised to continue its upward trajectory.

Figure 1 shows the trend in nominal hourly wages for the median worker (i.e., the worker in the middle of the wage scale). Nominal median wages accelerated in 1995-96, as productivity growth sped up and unemployment began to fall. Growth in median wages paused in 1998, and then, as the unemployment rate began to fall, accelerated again, growing more quickly than they had since the early 1980s. By the beginning of 2001, nominal median wages grew at an annual rate of 5%. The last few quarters, however, show a clear flattening of this trend, suggesting that the loosening of the labor market is affecting the pace of nominal wage growth.

fig 1

Figures 2A and 2B focus on the inflation-adjusted hourly earnings of low-, middle-, and high-wage workers. For men (Figure 2A), real wage growth accelerated between 1995 and 1998 as the unemployment rate headed toward historic lows. The tight labor market that developed over this period was particularly beneficial to low-wage male workers, whose annualized real earnings growth rose from 1% in 1995 to 3% by the end of 1998, the highest recorded rate since this series began in 1979:1. Note that growth at the 10th percentile for males kept pace with the growth rate for high-wage males, leading to a reduction in the wage gap between these earners over this period.

fig 2a

fig 2b

Wage growth for females (Figure 2B) followed a similar pattern. In fact, growth in the real hourly earnings of low-wage women far outpaced that of middle- and high-wage women in 1997 and through the first half of 1998.

At this point, it is unclear what led to the post-1998 decline in the real growth of low-wages. Faster inflation provides part of the explanation, particularly in late 1999 and early 2000, as energy prices spiked. But the nominal wage growth of low-wage workers also slowed, even though unemployment continued to fall. For low-wage females-whose wages are closely tied to that of the real minimum-inflation may have played a role, eroding the real value of the minimum wage by 5% from 1998 to 2000. Inflation, however, does not explain the decline for low-wage males, whose 10th percentile wage was about $6.50 in 1998, and thus less likely to be moved by the real decline in the minimum wage (which is $5.15). It is possible that the continued increase in the supply of low-wage labor generated by welfare reform played a role in dampening wage growth over this period (though this clearly was not the case in the earlier years of welfare reform).

Despite the recession, real wages continue to rise for all workers. But wages are not immune to the economy’s slowdown, as the flattening in growth for low-wage workers makes clear. Note that by the end of 2001:3, the old pattern of growing inequality had returned, with the earnings of high-wage workers growing faster then those of middle- and low-wage earners.


1. The underlying data represent the seasonally adjusted trend component. We use the trend to avoid the statistical “noise” present in these quarterly wage series. As discussed in our online QWES data appendix, the trend includes a stochastic component, so that it is more likely to reflect turning points in the data than more commonly used trend filters.

For more information, see the data appendix.

Check out the archive for past QWES.

See more work by Jared Bernstein